Performance Prediction and Optimization of Low Pressure Steam Turbine Radial Diffuser at Design and Off-Design Conditions Using Streamline Curvature Method

Abstract

The performance of the radial diffuser of a low pressure (LP) steam turbine is important to the power output of the turbine. A reliable and robust prediction and optimization tool is desirable in industry for preliminary design and performance evaluation. This is particularly critical during the tendering phase of retrofit projects, which typically cover a wide range of original equipment manufacturer and other original equipment manufacturers designs. This work describes a fast and reliable numerical approach for the simulation of flow in the last stage and radial diffuser coupled with the exhaust hood. The numerical solver is based on a streamline curvature throughflow method and a geometry-modification treatment has been developed for off-design conditions, at which large-scale flow separation may occur in the diffuser domain causing convergence difficulty. To take into account the effect of tip leakage jet flow, a boundary layer solver is coupled with the throughflow calculation to predict flow separation on the diffuser lip. The performance of the downstream exhaust hood is modeled by a hood loss model (HLM) that accounts for various loss generations along the flow paths. Furthermore, the solver is implemented in an optimization process. Both the diffuser lip and hub profiles can be quickly optimized, together or separately, to improve the design in the early tender phase. 3D computational fluid dynamics (CFD) simulations are used to validate the solver and the optimization process. The results show that the current method predicts the diffuser/exhaust hood performance within good agreement with the CFD calculation and the optimized diffuser profile improves the diffuser recovery over the datum design. The tool provides General Electric the capability to rapidly optimize and customize retrofit diffusers for each customer considering different constraints.